Erkenn DOI 10.

1007/s10670-011-9304-0 ORIGINAL ARTICLE

Explaining Perceptual Entitlement

Nicholas Silins

Received: 9 February 2010 / Accepted: 11 June 2011 Springer Science+Business Media B.V. 2011

Abstract This paper evaluates the prospects of harnessing anti-individualism about the contents of perceptual states to give an account of the epistemology of perception, making special reference to Tyler Burges (2003) paper, Perceptual Entitlement. I start by clarifying what kind of warrant is provided by perceptual experience, and I go on to survey different ways one might explain the warrant provided by perceptual experience in terms of anti-individualist views about the individuation of perceptual states. I close by motivating accounts which instead give a more prominent role to consciousness.

1 Introduction Lets take it for granted that our perceptual states give us warrant for some of our beliefsroughly speaking, our perceptual states make it rationally permissible for us to hold some of our beliefs. Weve set the question of skepticism aside, but we still face the fundamental question of how our perceptual states give us warrant for beliefs. Burge (2003) answers the fundamental question by looking at what makes perceptual states represent the world in the way they do. The primary aim of my paper is to evaluate his important and under-discussed account. To begin to explain his account and why it is important, let me orient the account with respect to two further questions about how perceptual states warrant beliefs. The rst further question concerns reliability: (1) What role, if any, do considerations about reliability play when a perceptual state provides warrant?
N. Silins (&) Sage School of Philosophy, Cornell University, 218 Goldwin Smith Hall, Ithaca, NY 14853, USA e-mail: ns338@cornell.edu

123

N. Silins

Roughly speaking, when we address this question, we look outside experiences, at how they are correlated with states of affairs in the world. On an extreme reliabilist view, a perceptual state gives one warrant to believe that p if and only if it is reliably correlated with its being the case that p. Some consequences of this view are too demanding. The view predicts that, if an evil demon made your experiences misleading most of the time, they would cease to give you warrant for ordinary beliefs. Other consequences of the view are not demanding enough. Our experiences are reliably correlated with our brain states, but our experiences do not give us warrant for beliefs about which brain states we are in. To see a very different approach from the extreme reliabilist approach, consider the following further question about consciousness: (2) What role does the conscious character of a perceptual state play, if any, when a perceptual state provides warrant? Roughly speaking, when we address this question, we look inside experiences, at what it is like to have them. On an extreme internalist view, a perceptual states giving one warrant to believe that p is explained entirely by what it is like to have the experience, so that conscious duplicates are given warrant to believe exactly the same things by their experiences. This view also faces a challenge: it needs to explain how it stands with respect to the idea that a warrant for a belief is a good guide to the truth of the belief. The extreme internalist view needs either (1) to provide a good sense in which warrants are truth-conducive (compatible with the retention of warrant if one becomes radically deceived), or (2) convince us to reject the idea that warrants must be truth-conducive. Tyler Burges landmark 2003 paper, Perceptual Entitlement, defends a nuanced view which answers both of the questions we have highlightedand otherswhile avoiding the extreme internalist position as well as the extreme reliabilist position. The paper stands out for the way in which it assigns warrant the role of providing a good route to truth, while avoiding the difculties of extreme reliabilist positions. In particular, the paper develops a novel application of antiindividualism about perceptual representation, which holds that perceptual states have their representational contents in virtue of relations to the environment. The resulting account is striking, not least because it entails that visual consciousness is actually not required for the acquisition of visual perceptual warrant. On this account the conscious character of a perceptual state need not play any role when the perceptual state provides warrant. Burges account is a special case of a more general approach in the epistemology of perceptual belief. This approach emphasizes, in one way or the other, the proper functioning of perceptual or cognitive systems (Plantinga 1993, 1996; Bergmann 2006). Particular instances of the approach go on to vary in their details (and even their aims: what Plantinga calls warrant is quite different from what Burge calls warrant). Burges account is important to single out as a case study for evaluation of the proper function approach. First, Burge defends his account largely by invoking

123

Explaining Perceptual Entitlement

principles rather than by rening hypotheses in light of increasingly complicated counterexamples. Considering his defense thus promises to provide a deeper understanding of what motivates the proper function approach. Second, Burges account stands out because it does not use the controversial notion of design (although see Bergmann 2006: 144146 for helpful discussion of design). Finally, Burges account has attracted vastly less critical scrutiny in the literature, indeed none so far as I am aware (it has however inuenced similar accounts given by Peacocke 2004 and Sawyer and Majors 2005). In this paper I will evaluate Burges position while bearing in mind the more general prospects of the proper function approach. Although I will not develop an alternative account here, I should say that I have a broader agenda. The agenda is to block Burges challenge to views which assign the conscious character of perceptual states a much more prominent role. Examples of such views can be found in the work of Pollock (1974), Pryor (2000, 2004), Huemer (2001, 2007), and Silins (2008, forthcoming). My focus will be on replying to Burges objections against them. Let me now outline the rest of the paper. In Sect. 2, I will address Burges account of the kind of warrant perceptual states characteristically provide. Here I will clarify his inuential distinction between entitlement and justication, and I will bring out a (surmountable) problem for arguments which use the distinction. In Sect. 3, I will turn to Burges account of how perceptual states provide warrant. The core of the account, to be explained in much more detail in Sect. 2, is that (1) perceptual states must be reliable in a certain non-accidental way to provide entitlement, where (2) anti-individualism explains how perceptual states are reliable in the relevant non-accidental way. I will argue that the rst point is not established. I will also argue that, even if the rst point is granted, one need not invoke anti-individualism to provide an equally good explanation of the non-accidental reliability of perceptual states. Finally, in the concluding sections of the paper, I will bring out an advantage of views which assign an epistemic role to consciousness.

2 Burges Account of the Kind of Warrant Provided by Perceptual States Burges focus is on how perceptual states provide a kind of warrant he calls entitlement. We can understand what entitlement is by understanding the distinction he draws between entitlement and justication: Entitlement is epistemically externalist inasmuch as it is warrant that need not be fully conceptually accessible, even on reection, to the warranted individual The other primary sub-species of epistemic warrant is justication. Justication is warrant by reason that is conceptually accessible on reection to the warranted individual (504). The distinction is useful, but it will take some work to pin it down. The complication is that there are varying degrees to which a subspecies of warrant may

123

N. Silins

be externalist or internalist.1 This complication has a major impact on how to argue that perceptual states provide entitlement. In order to gloss the entitlement/justication distinction, we need to consider several grades of access to warrants.2 First, it is one thing to access a state which is a warranting state, another to access the fact that it is a warranting state: (Access1): One has conceptual access on reection to a perceptual state which provides one with warrant. (Access2): One has conceptual access on reection (1) to a perceptual state which provides one with warrant, and (2) to the fact that it provides one with warrant. Second, it is one thing to access properties of a state which make it a warranting state, and another to access the fact that those properties play that role: (Access3): One has conceptual access on reection (1) to a state which provides one with warrant, and (2) to the fact that it provides one with warrant, and (3) to some or all of the features of the perceptual state which make it provide one with warrant. (Access4): One has conceptual access on reection (1) to a state which provides one with warrant, and (2) to the fact that it provides one with warrant, and (3) to some or all of the features of the perceptual state which make it provide one with warrant, (4) the fact that those features make the state provide one with warrant. For each of these grades of access, one has conceptual access on reection to a state (or fact) just in case one is in a position to warrantedly believe that one is in the state (or that the fact obtains). Our key questions now concern the relation of entitlement and justication to the Access conditions highlighted above. Must one satisfy any of them to have entitlement? Must one satisfy all of them to have justication? We can make headway by considering how Burge argues for the view that perceptual states provide entitlement: The claim that reason or justication is the only sort of epistemic warrant can be seen as a stipulative restriction on what warrant is to mean. But if it is a substantive claim, it hyper-intellectualizes epistemology Children and higher non-human animals do not have reasons for their perceptual beliefs. They lack concepts like reliable, normal condition, perceptual state, individuation, defeating condition, that are necessary for having such reasons. Yet they have perceptual beliefs There are legitimate questions about animals and young childrens entitlement to their perceptual beliefs (5289). The argument uses the requirement that, to have reasons or justication for perceptual beliefs, one must have concepts like reliable, individuation, defeating

The senses of internalist and externalist in play concern accessibility. For other senses in which the terms are used in epistemology, see Pryor (2001) or Wedgwood (2002). Here I draw on Alstons classic work on varieties of access externalism (see his 1989, essay 9).

123

Explaining Perceptual Entitlement

condition, and perceptual state.3 The idea is presumably that one must have concepts in some way similar to the enumerated concepts to have reasons or justication, where this is not yet to say that one must have all of the enumerated concepts. Given how sophisticated the concept of individuation is, a notion of justication which required possession of that concept would be too demanding to apply in very many cases. Given the kind of concepts required to have reasons or justication, there are high internalist standards for one to have reasons or justication: one must at least satisfy Access1 and Access2. Since our main focus will be on entitlement, I leave open exactly how strong the access requirement is for justication.4 Lets now turn to entitlement. Since children and animals have entitlements for perceptual beliefs, yet do not have all the concepts required to satisfy any of the Access conditions, it is not necessary to satisfy any Access condition to have perceptual entitlement. We need a sufcient condition for entitlement as well. I assume that, if one does have warrant from a perceptual state while failing to satisfy Access conditions 1 through 4, then one does have entitlement from the state. This is not to say that one must fail to satisfy Access1 through Access4 to have perceptual entitlement. One may have entitlement from a perceptual state despite satisfying the Access conditions, so long as one enjoys warrant from the state such that one can have it without satisfying the Access conditions. It is thus possible to start out with entitlement for a belief, and then to acquire concepts which allow you to have justication for the belief, without losing your entitlement for the belief during the process. The entitlement/justication distinction is therefore not exclusive. To explain the entitlement/justication distinction, we still need to address an important choice point. On one way of proceeding, the distinction will straightforwardly be exhaustive. Justication will be warrant such that one needs to satisfy all of Access1 through Accessn to have it, and entitlement will be warrant such that one need not satisfy all
3

For ease of exposition, Ill assume in the main text that talk of reasons is interchangeable with talk of justication. As Tyler Bure pointed out to me, however, there arguably are reasons which arent justications. In the inferential case, a subject might transition from a belief in a premise to a belief in a conclusion without any grasp of the relevant rule of inference. In such a case the subject might have a reason for her belief in the conclusion without yet meeting the standards for having a justication for her belief in the conclusion. For further evidence that there are high standards for justication or reasons, consider the following passage: Understanding the warranting force of being in a relevant-seeming perceptual state and having some grip on the general shape of conditions that undermine such force are part of having a reason (529). A complication is that, on Burges view, it may be that perceptual states strictly speaking never give us justication for beliefs. He seems to hold that only entities which are propositional in some suitable sense provide justication, where perceptual states are not propositional in this sense (see section V of his paper). This complication can be nessed. We may simply consider what it takes for the proposition that one has a certain perceptual state to give one justication or be a reason for a perceptual belief. As far as the formulation of the Access conditions is concerned, we may revise them along the following lines: (Access1): When a source of warrant for one is that one is in perceptual state P, one is in a position to warrantedly believe that one is in P. (Access2): One is in a position to warrantedly believe that (1) one is in perceptual state P and (2) a source of warrant for one is that one is in P.

123

N. Silins

of Access1 through Accessn to have it. I actually wont follow this routeit fails to meet a crucial desideratum. The concept of entitlement is intended to be epistemically externalist. Given that the concept of justication is a strongly internalist concept, we will obtain an exhaustive distinction only if entitlement is introduced as a concept which is not strongly internalist. However, a concept which is not strongly internalist can simply be weakly internalist (by merely demanding the satisfaction of Access1). If we construe entitlements as any warrants which fail to be strongly internalist, then, the claim that there are entitlements will leave open whether all warrants have access requirements. The claim that there are entitlements will leave open whether all warrants require the satisfaction of at least Access1. But we want the claim that there are entitlements to close off the possibility that all warrants have access requirements. To obtain a properly externalist concept of entitlement, we need to build into the concept that there are no access requirements for one to satisfy it. Entitlements should be dened as warrants that one can have without satisfying any of the Access conditions 14. In sum, there are strong internalist access requirements for one to have justication, and there are no internalist access requirements for one to have entitlement.5 There is an important upshot of the way we need to distinguish entitlement and justication. The distinction is not exhaustive: there are subspecies of warrant which consist neither of entitlement nor of justication. In particular, there is a subspecies of warrant W such that one has it only if one satises Access1, although one need not satisfy Access2 to have W. Compare a creature who does have minimal access to its warranting perceptual state and a creature who has a warranting perceptual state without any access whatsoever to it. This is not a gerrymandered difference like the difference between a warranting perceptual state which occurs in California and a warranting perceptual state which occurs in New York. This is a difference in epistemic terms, a difference in warrant. So the entitlement/justication distinction is not exhaustive. We now have a lesson about how to argue that there are perceptual entitlements.6 One tempting way to argue is roughly as follows: (A) Children and animals have warranted perceptual beliefs. (B) Children and animals do not have justications for perceptual beliefs. So, (C) Children and animals have entitlements for perceptual beliefs.

This argument is invalid. Given that the entitlement/justication distinction is best understood as being non-exhaustive, someone could have warrant, lack justication, and still lack entitlement as well.
5 6

For different takes on the entitlement/justication distinction, see Casullo (2007) or Neta (2010).

The point also affects Burges 1993 argument that testimony is a source of a priori entitlement, which seems to require that the entitlement/justication distinction is exhaustive. For further discussion of that argument see Malmgren (2006, Section 2).

123

Explaining Perceptual Entitlement

To establish that perceptual states provide entitlement, we should provide cases in which perceptual states provide warrant to a subject while the subject fails to satisfy any of Access1 through Access4. Children and animals do provide such cases. So perceptual states do provide entitlement. Further, since one does not lose perceptual entitlement by gaining conceptual sophistication, even conceptually sophisticated adults will enjoy perceptual entitlements as well.7 Now that we have explained and defended the claim that perceptual states provide entitlement, we should consider an important further claim Burge makes about the kind of warrant provided by perceptual states: Perceptual beliefs commonly do not rest for their epistemic warrant on any other beliefs (526, n.19) We can call this thesis the Basicness thesis. It will be crucial to the defense of further components of Burges overall position. The claim is that (Basicness): Perceptual states sometimes give one warrant for perceptual beliefs in a way which does not rely on warranted background beliefs one has. The Basicness thesis does allow that background beliefs can interfere with the warrant one has from a perceptual state. For example, if one has reason to believe that one is seeing a mirage, ones perceptual warrant will be undermined. However, according to the Basicness thesis, perceptual states do not give one warrant in virtue of ones having a warranted belief that one is not seeing a mirage. More generally, perceptual states give one warrant independently of warranted background beliefs. I take Burges defense of the Basicness thesis to be similar to his defense of the claim that perceptual states provide entitlement (see p. 544). The idea is that the denial of the Basicness thesis would be an over-intellectualized epistemology of perception. If the Basicness thesis is false, then animals and children lack warranted perceptual beliefs, given that they do not have the sorts of beliefs which would certify their perceptual states as reliable or warrant-conferring. Since animals and children do have warranted perceptual beliefs, the Basicness thesis is true.

3 Burges Account of How Perceptual States Provide Entitlement 3.1 The Exposition of Burges Account We have seen how to understand Burges helpful distinction between entitlement and justication, and how to set out a valid argument for the conclusion that perceptual states provide basic entitlement. This conclusion is attractive and

To rescue the ABC argument for perceptual entitlement, one might formulate the entitlement/ justication distinction so that it is exhaustive. But there is no reason to try to rehabilitate the ABC argument, given that the conclusion can be reached equally well in other ways.

123

N. Silins

important (although it may not be the most important thesis in the vicinity).8 The Basicness thesis is not yet distinctive of Burges position however, since it is in effect endorsed by philosophers such as Pollock (1974), Pryor (2000, 2004) or Huemer (2001, 2007). Lets now consider Burges distinctive position about how perceptual states provide entitlement. The story is complex. Ill set out its main components before discussing how to argue for the position. Burge highlights two key aspects of perceptual states to account for their role in entitlement. The rst is their anti-individualist nature. The second is their reliability. Lets take each in turn. The role of anti-individualism in the account is to connect the natures of perceptual states with veridical representation. This component of Burges position is inuential, having been taken up in different ways by Peacocke (2004) and Sawyer and Majors (2005).9 According to Burge, perceptual states have the contents they have in virtue of causal interactions with their referents and accurate representation of their referents. This claim will need to be qualied in what follows, but we can rely on a rough grip on it for now. The crucial epistemic application for now is that Reliance on perceptual states is warranted partly because the very identity of the states is constitutively and explanatorily associated with veridical representation (532). To see the point, compare perceptual states with states of imagination. States of imagination do not give one entitlement to endorse what they represent. A promising explanation of why is that states of imagination are not individuated by veridical representation on their part (this allows that states of imagination are individuated in some other anti-individualist way, say by their association with perceptual states which are individuated by the veridical representation done by the perceptual states). If perceptual states are individuated by veridical representation, it is reasonable to expect this fact to account at least partly for how perceptual states give entitlement, in contrast to states of imagination.
8

The Basicness thesis formulated by Burge is weaker than the following claim: (Strong Basicness): Perceptual states sometimes give one warrant for perceptual beliefs in a way which does not rely on ones having warrant in favor of some background proposition. If someone denies Basicness, they must explain the capacity of an experience to warrant belief in terms of a persons actually having warranted background beliefs. Such a view is indeed psychologically demanding. If someone denies Strong Basicness, they need not make the same psychological demands. Their focus is instead on ones being in a good epistemic position with respect to certain propositions, whether or not one actually has the beliefs or perhaps even has the concepts required to have them. I believe that Strong Basicness is the most important thesis in the ballpark of the Basicness thesis. It is of a piece with Pryor (2000, 2004)s claim that experiences give immediate justication for beliefs about the external world. The Strong Basicness thesis is not established by the arguments in terms of concept possession we have seen. Having noted the difference between Basicness and Strong Basicness, Ill mostly set it aside in what follows, although I do return to it in note 12. For a quite different application of anti-individualism in the epistemology of perception, see Brewer (1999).

123

Explaining Perceptual Entitlement

The role of anti-individualism in the account is importantly not to guarantee that perceptual states are reliable. The representation of danger by a rabbit, for instance, need not be reliable, even though it must occur in the background of at least some success (517). Anti-individualist individuation is thus not sufcient for reliability, and the reliability condition is a further condition in addition to the anti-individualist condition. Burge states it as follows: To contribute to epistemic entitlement, a perceptual state (type) must be reliably veridical in the perceptual systems normal environment. The normal environment is the one by reference to which the perceptual content of the perceptual state is explained and established When a perceptual state is reliable in the normal environment, and certain other conditions on entitlement are met, the individual is entitled in any environment to perceptual beliefs that are appropriately derived from the perceptual state (532). There are two important claims here. The rst is a necessary condition for perceptual entitlement: (Necessity): A perceptual state (type) provides entitlement only if it is reliable in the environment in which its content is xed. The second claim is in effect a sufcient condition for perceptual entitlement: (Sufciency): If a perceptual state (type) is reliable in the environment in which its content is xed, it will defeasibly provide entitlement in any environment. My formulation speaks of ones having perceptual entitlement defeasibly, without yet telling us what can defeat the entitlement. I wont spell out what can get in the way, but see Burges section VII and n. 24 for discussion. The key point for now is that, if the intended version of Sufciency is true, then a perceptual state can provide entitlement even if it comes to be unreliable. For example, the food represented by the state might come to be largely replaced by perceptually indistinguishable dust (539), or the bearer of the state type might come to be deceived by an evil demon. According to Burge, the entitlement provided by a perceptual state can travel to such new environments, even if the perceptual state is not reliable in those environments. There is an extremely important further consequence of Sufciency: the conscious character of perceptual states need not play any role in perceptual entitlement. What is crucial is the individuation and reliability of the perceptual state, leaving aside what it is like to be in the state. The picture thus allows that a suitable blind-sighted subject can be as warranted in their perceptual beliefs as we are warranted in our perceptual beliefs. As long as her perceptual states are individuated and reliable in the right way, without any defeaters in play, her perceptual beliefs will be warranted, despite her lack of visual consciousness.10 Notice that the striking claim stands whether or not the conscious character of a state is itself individuated as a relation to content (as is maintained by intentionalists
10 One could say that the absence of visual consciousness is itself always a defeater of perceptual entitlement, but nothing in Burges account indicates he would say this.

123

N. Silins

such as Tye 2000 or Byrne 2001). To make the point, we consider perceptual states which have no conscious character, even though they do meet the sufcient conditions specied by Burges account. We need not assume that the blindsighted creatures states represent the world as being the same way as it is according to our conscious perceptual states, all that matters is the way in which the blindsighters perceptual states are reliable. Burge also makes an important further claim about the role of reliability in entitlement: (The Grounding Requirement): If the reliability of a perceptual representation is not grounded in the individuation and nature of the state, then the reliability cannot yield entitlement (532). Lets use the term incidental reliability for the reliability of a perceptual state which is not grounded in its nature. If the grounding requirement obtains, the incidental reliability of a perceptual state is entirely irrelevant to entitlement in addition to being insufcient for entitlement. These claims are different. For example, one could think that non-incidental reliability is indeed necessary for entitlement, while still allowing that an increase of incidental reliability can enhance the entitlement one obtains from a perceptual state. According to Burge, however, incidental reliability has no impact at all on the entitlement conferred by a state.11 One might naturally wonder what it means for the reliability of a perceptual state to be grounded in its nature. This question is pressing since, on Burges view, it is possible for a perceptual state to represent a property without being reliable in attributing that property. Recall his earlier separation of the reliability condition for entitlement from the individuation condition for entitlement, and the example of the representation of danger by a rabbit. The reliability of a perceptual state in representing a property is not necessitated by its nature. The idea is not to include any old reliability had by a perceptual state in the environment in which its content is xed. Suppose a perceptual state comes to represent a property thanks to some successful interactions with the property, even though the overall reliability of the perceptual state is secured thanks only to the interventions of a guardian angel, who often alters the environment to accord with the way it is represented as being. Such reliability is not grounded in the individuation of the state. To contribute to entitlement, the reliability of the perceptual state presumably must play a role in maintaining the content of the perceptual state. This is not to say
Related claims are made in the following passages: If the reliability of a perceptual state is not inductively learned by the individual, and is not grounded in the individuation and nature of the state, then reliably veridical perceptual states make no contribution to entitlement (533) Both unreliability and reliability in conditions other than those that played a role in explaining the nature of the perceptual state and the exercise of the perceptual competence are accidental relative to those natures. So reliability and unreliability in such conditions are irrelevant to the connection between warrant and veridicality (536) The only reliability that is relevant to the contribution of perceptual states to perceptual warrant is one that attaches to the states normal functioning in the conditions that explain their natures. That is the only reliability that is non-accidental relative to the natures of the perceptual states (536).
11

123

Explaining Perceptual Entitlement

that reliability is a necessary feature of the perceptual state, but rather to say that the reliability of the state plays an explanatory role with respect to its content. An advantage of this construal is that it allows for a perceptual state to initially fail to provide entitlement, due to its unreliability when its content is rst xed, although later come to provide entitlement, thanks to subsequent reliability which plays a role in maintaining its content. Although more could be said to clarify the grounding requirement, in what follows I will assume that we have a working understanding of the claim. This completes my exposition of Burges account. 3.2 The Defense of Burges Account What considerations are there in favor of the account? One question here is whether there should be any requirement for entitlement in terms of reliability. I will set this question aside, however, and focus on the defense of Burges more distinctive claims. My aim is to address the account on its own terms. I will discuss two main arguments. The rst argument relies on the Basicness thesis, the claim that perceptual states provide one with entitlement independently of ones background beliefs. I nd the argument here: Having a reliable perceptual competence with respect to a particular perceptual state makes that state, through its nature, part of a good route to truth. This point reects the fact that the occurrence of a percept, functioning normally within a cognitive system, in itself makes a defeasible contribution toward ones being entitled to a belief that is appropriately related to it (534535). I take this argument to depart from the background claim that, if perceptual state give us entitlement, then they give us basic entitlement. The rest of the argument focuses on what it takes for perceptual states to give us basic entitlement. First (and trivially), if perceptual states give us basic entitlement, then they give us entitlement in a way which does not rely on warranted background beliefs we have. Second (and substantively), if perceptual states do not rely on our warranted background beliefs to give us entitlement, then they do rely on their own natures to give us entitlement. The percept must in itself make a contribution to entitlement, that is, its non-incidental properties must explain how it is able to provide entitlement. Third, if some feature of a perceptual state explains how it provides entitlement, that feature has to do with the reliability of the state. Therefore, if perceptual states give us basic entitlement, they are non-incidentally reliable. Now since perceptual states give us entitlement only if they give us basic entitlement, it follows that they give us entitlement only if they are non-incidentally reliable. The problem with the argument is that it reads too much into the thought that perceptual states give us basic entitlement. If perceptual states give us basic entitlement, it indeed follows that they give us entitlement in a way that does not rely on our warranted background beliefs. However, the Basicness claim merely gives us negative information about how

123

N. Silins

perceptual states give us entitlement, saying that background beliefs need not play a certain role. The Basicness claim does not give us positive information about how perceptual states do give us entitlement. It denies a claim about what is necessary for perceptual states to give entitlement, without giving any information about what is sufcient for perceptual states to give entitlement. In particular, Basicness leaves open various roles reliability could play in entitlement. For example, the claim leaves open that entitlement could fail to travel to environments in which a perceptual state type comes to be unreliable. It is therefore no surprise that the thesis is accepted by epistemologists who emphasize incidental reliability, such as Goldman (2008), and by epistemologists who do not emphasize any kind of reliability, such as Pryor or Huemer. It is far from being the key to the epistemology of perception.12 I now turn to another line of argument for Burges account. It is an inference to the best explanation: Reliable connections to the world that are accidental relative to the conditions that individuate the individuals perceptual states make no contribution to epistemic entitlement. This fundamental feature of warrant is underwritten by the connection between perceptual entitlement and the anti-individualist natures of perceptual states. That connection explains the reliable success of those perceptual states that are reliable, in terms of the environmental conditions and environment-state relations that determine the natures of those states (535). This line of reasoning departs from the grounding requirementthe claim that, if the reliability of a perceptual state is to play a role in explaining perceptual entitlement, that reliability must be grounded in the individuation of the state. The focus is on how best to explain the satisfaction of the requirement by perceptual states. The key move is to say that anti-individualism with respect to perceptual contents provides the best explanation of how the reliability of a perceptual state can be grounded in its nature. This move promises to establish the Necessity component of Burges position, as well as the recruitment of anti-individualism to give an account of perceptual entitlement. Similar reasoning seems intended to establish the Sufciency component of Burges position: The only reliability that is relevant to the contribution of perceptual states to perceptual warrant is one that attaches to the states normal functioning in the conditions that explain their natures. That is the only reliability that is nonaccidental relative to the natures of the perceptual states. Thus the normal
12 A further problem arises because of the weakness of Burges Basicness thesis. As discussed in note 8, the thesis could be true even if a role is played in entitlement by ones being well-positioned with respect to certain propositions. The Basicness thesis rules out that all perceptual warrant relies on warranted background beliefs one actually has, it does not rule out that all perceptual warrant relies on ones being epistemically well-positioned with respect to background beliefs. This problem gets in the way of using the weak version of Basicness to establish Sufciency. If the weak version of Basicness is true, your perceptual state could satisfy the antecedent of Sufciency, while you still fail to have perceptual entitlement, because of a failure to be well-positioned to hold crucial background beliefs.

123

Explaining Perceptual Entitlement

environmentthe conditions in which content is explained and established is privileged in explanations of entitlement This privilege entails the irrelevance to entitlement of reliability and unreliability in other conditions. So it extends to indiscernible abnormal environments that the individual might contingently nd himself in (536). Let me now evaluate the inference to the best explanation. One natural question concerns whether we should accept the grounding requirement. In what follows, however, I will simply assume that the grounding requirement is true, and investigate other problems which arise. For Burges inference to the best explanation to succeed, rival explanations need to be blocked. I will argue that this crucial work has not been done. There are two quite different explanations which need to be addressed. Each of them develops a different way of satisfying the grounding requirement. When we consider these rival accounts, we will expand our sense of the space open to a broadly proper function approach in the epistemology of perception. When we evaluate them, we will see an important advantage of the quite different approach of emphasizing consciousness rather than proper function. The rst rival account extends Burges use of anti-individualism. His focus was on the reliability of a perceptual state which plays a role in xing and maintaining the content of the perceptual state. One could go further and focus on the reliability a perceptual state has as a matter of its essence. On this line of thought, perceptual states are reliable as a matter of their nature, in the very strong sense that they are reliable in any environment in which they can be found. The essentialist alternative to Burges account must be formulated with care. When ones focus is on how perceptual states have the contents they do, there is only so much reliability one can squeeze out of their essence. Burge makes the point as follows: One can erroneously perceive an instance of a color shade or shape type as being present, even though no individual has ever interacted with such an instance. This is possible as long as an ability to represent a spectrum of color shades or shapes is in place. Instances of other types in the spectrum must have been interacted with (531, n. 23). Let me focus on the case of color and set other cases aside. The point here is not just that one can have a perceptual representation of a missing shade of blue, where that shade of blue is nowhere instantiated in ones environment. The point here is more general: for any perceptual representation of any shade, one could have perceptual representations of that shade which were entirely unreliable, provided that enough of ones representations of other shades were reliable. No particular shade is the anchor of all color representation. One therefore better not develop the essentialist alternative as follows: (Essentialist Reliability, First Pass): A perceptual representation P of a shade provides entitlement for a belief attributing that shade only if P is essentially a reliable representation of that shade.

123

N. Silins

Such a position would have skeptical consequences, given that For any perceptual representation P of any shade S, P is not essentially a reliable representation of S, since it is possible for P to occur without being a reliable representation of S. To avoid the difculty, one should say that it is of the essence of a representation of a shade to occur in a network of reliable representations of other shades. On this line of thought, color representation cannot get off the ground unless enough representations of shades are reliable (or at least have been reliable). A corresponding adjustment can be made to the reliability requirement for entitlement. On this line of thought, your perceptual representations of the missing shade of blue can provide entitlement for beliefs attributing the missing shade of blue, even if it is not of their essence to reliably represent the missing shade of blue. If it is of their essence to be embedded in a network of (at least previously) reliable representations of other shades, they will be able to provide entitlement. Lets then state the alternative requirement as follows: (Essentialist Reliability): Necessarily, if a perceptual state type gives one entitlement, then it is of the essence of the state type to be holistically reliable. This formulation leaves open exactly how to cash out holistic reliability. However, the proposal does provide a demanding requirement for perceptual entitlement in terms of reliability, arguably without demanding too much. I have focused on a requirement for entitlement, but one can rewrite Burges overall account of perceptual entitlement in terms of essentialist reliability. Such an account is plausibly at least as good as his own. In particular, there is no doubt that the account forges a connection between the way a perceptual state is individuated and its reliability. Given that the account focuses on the essence of a perceptual state, the account proposes a tighter connection than Burge. So it can arguably provide a better explanation of how perceptual states satisfy the grounding requirement for reliability. Given the focus on essence, the account also might be in a better position to explain Sufciency, the ability of perceptual entitlement to travel to environments in which most or all occurrences of a perceptual state come to be non-veridical. Finally, given the focus on holistic reliability, the account arguably gives a better treatment of how we could have entitlement when we seem to see a shade of blue which is in fact a missing shade of blue. Given these virtues, it is hard to see how Burge could block the rival account, yet he must do so for his inference to the best explanation to succeed. There is an important further alternative which sets aside the content of perceptual states. This alternative focuses on the attitude involved in a perceptual state. By evaluating this alternative, we will expose a difculty for applying antiindividualism with respect to content in the epistemology of perception, and we will reveal an important advantage of approaches which highlight consciousness instead. First consider the view on which, when one has perceptual entitlement to believe that a is F, one has the entitlement in virtue of being in the state of seeing as Fness (Johnston 2006). As a matter of the essence of the state of seeing as Fness, one sees as Fness only if a is F. So the state of seeing as Fness is reliable of its very nature,

123

Explaining Perceptual Entitlement

there is no question of its failing to satisfy the grounding requirement. However, on this picture, when we are explaining our entitlement to beliefs of the form (a is F), we need not invoke anti-individualism about perceptual content. We instead focus on the individuation of perceptual attitudes. This explanation is a clear rival to Burges explanation. The crucial question is whether Burges explanation is better. I will address the debate between them in a moment. Right now let me emphasize that, from the perspective of the grounding requirement alone, states of seeing as Fness should seem like excellent candidates to be sources of perceptual entitlement. A further challenge here is that perceptual states might meet the grounding requirement without even having externally individuated content. Here we should consider the factive state of seeing that p, where as a matter of necessity (and essence) one sees that p only if p. Here again we have a perceptual state which is reliable of its very nature. Thus, if one held that only states of seeing that p provide perceptual entitlement, one would again respect the grounding requirement (see McDowell 1995 or Jackson 2011 for such a view). Importantly, states of seeing that p meet the grounding requirement whether or not they have externally individuated content. If one sees that p, where the content that p is determined in an individualistic way, one still remains in a state that is of its essence reliable with respect to the content that p. Here one could deny that a perceptual state has externally individuated content, while still providing an explanation of how it satises the grounding requirement. Such an account would be an especially important rival to Burges, since it abandons his emphasis on content. Lets call the family of views just outlined factivist views. They deserve this name because, on these views, you have perceptual entitlement to believe that p only if it is the case that p. They are therefore extremely demanding views about what it takes for one to have perceptual entitlement. However, they could allow for say inductive entitlement to believe that p when its not the case that p (as does Jackson 2011). So it is no objection to say that the very idea of entitlement allows for the possibility of having entitlement for a mistake. There is no doubt that Burge rejects the factivist views. He holds that It is a fundamental feature of perceptual warrant, hence perceptual entitlement, that it allows that an individual could be fooled while retaining warrant (537)13 The crucial question concerns how Burge will argue against the factivist views. It wont do to appeal to appeal to the grounding requirement, since the factivist views are entirely at home with such claims. Something further is required. One move is to appeal to the (strong) internalist view that, if two thinkers are phenomenally the same, then the perceptual beliefs of the thinkers are equally warranted. So long as two thinkers can be phenomenally the same while differing with respect to seeing that p, or seeing as Fness, the factivist views will be false. Burge however is no position to make this argument. For one thing, his overall view
A very minor point: what goes for perceptual warrant need not go for perceptual entitlement. For instance, perceptual warrant has the fundamental feature that one can have perceptual warrant by having perceptual justication as opposed to perceptual entitlement. However, perceptual entitlement is not itself such that one can have it by having perceptual justication as opposed to perceptual entitlement.
13

123

N. Silins

seems to be inconsistent with the internalist view in question. His overall view seems to allow that (1) in some cases of hallucination, one lacks perceptual entitlement and (2) such hallucinations have the same conscious character as nonhallucinations which do provide perceptual entitlement (note 24). For another, Burge eschews the use of the intuitions which are typically used to motivate internalist views. This not a mere ad hominem point. Burge indeed should avoid appeal to standard internalist intuitions, given that they will tend to oppose the broadly reliabilist approach he takes.14 A different move would be to appeal to the Sufciency claim formulated above. If this claim is true, then one could have perceptual entitlement for a false belief, thanks to being in a perceptual state which is individuated in a suitably reliable way without going all the way to being factive. The problem here is that the Basicness claim does not support the Sufciency claim, and we havent yet seen other reasons to believe the Sufciency claim. I reject the factivist views, for standard internalist reasons. However, I dont see what resources Burge himself has to reject the factivist views, given his other commitments and his other arguments. In sum, Burges inference to the best explanation will not succeed until these important rival explanations are addressed. 3.3 A Further Problem for Burges Account The factivist views enjoy at least one advantage over Burges overall position: Burge faces a challenge they do not. Posing the challenge will reveal an advantage of any view which emphasizes consciousness more than Burge does, without going to the factivist extreme. I take it that perceptual beliefs do not warrant themselves. The challenge for Burges position is to explain why perceptual beliefs are not self-warranting. This challenge arises because of the way in which basic perceptual beliefs are individuated (where a basic perceptual belief is understood to be a belief which takes a perceptual state at face value, without going beyond the content of the perceptual state). On Burges picture, basic perceptual beliefs are individuated as the conceptualizations of objects and properties represented by perceptual states. We can set aside what the conceptualization in question amounts to. Whats crucial here is that, since perceptual beliefs are individuated by corresponding perceptual states, if perceptual states are reliable in a way that is grounded in their individuation, then the corresponding perceptual beliefs will likewise be reliable in a way that is grounded in their individuation. As Burge puts the point: The perceptual state is individuated partly in terms of relations to kinds in the world that gure in successful perceptual representation. The concept is individuated through its relation to the percept, and ultimately through its relation to the empirical world (545).
14 I do not want to rely on such [internalist] intuitions. I think the entitlement in ordinary cases does not depend on how the individual does, but on the natures of his subsystems and states, and on the connection of these natures to the objects of perception (note 24 p. 537).

123

Explaining Perceptual Entitlement

On Burges overall picture, it then wont be clear why perceptual beliefs are incapable of warranting themselves. Thats because a perceptual belief with the content that p will have the sort of feature which explains the ability of a perceptual state to provide entitlement for a belief with the content that p. In order to explain entitlement, after all, Burge focused on reliability which is suitably grounded in the individuation of contentsif a state has a content which is reliable in a suitable way, the state defeasibly gives one entitlement to endorse that content (or a conceptualization of that content). But this condition is satised by basic perceptual beliefs as well as by perceptual states, given that the individuation of basic perceptual beliefs is parasitic on the individuation of perceptual states. When a perceptual state is reliable in a non-incidental way, the corresponding perceptual belief will be reliable in a non-incidental way as well. The upshot is that we should expect basic perceptual beliefs to defeasibly give one entitlement to endorse their very own contents. Again, the perceptual state has a feature which explains its ability to give entitlement to believe that p, and this feature is possessed by the belief that p itself. So it looks like perceptual beliefs should be self-warranting. Given that a counterpart of Burges Sufciency claim holds for basic perceptual beliefs themselves, there is a challenge for him to explain why no perceptual beliefs warrant themselves. He could invoke some extra constraint on entitlement to meet the challenge, the question though concerns which constraint he will invoke, and why. There is no refutation here of his view, but there is a challenge for the view.15 This challenge will arise for any view which aims to harness anti-individualism to explain perceptual entitlement, as do Peacocke 2004 and Sawyer and Majors 2005. The factivist view faces no such challenge, even though it does face other challenges. Even if the content of perceptual beliefs is reliable in a way that parallels the reliability of the content of perceptual states, the individuation of the respective attitudes remains entirely different. So there is no threat that the factivist view will allow perceptual beliefs to warrant themselves.16 The challenge just outlined is of much wider importance. It illustrates an advantage of any view which looks beyond the individuation of perceptual content to explain perceptual entitlement. Such views need not be formulated in terms of factive attitudes, and arguably should not be formulated in terms of factive attitudes. Consider instead the conscious character of experience. What it is like to have a visual experience is different from what it is like to have a perceptual belief. So any view which emphasizes consciousness will avoid the challenge to explain why perceptual beliefs do not warrant themselves. One can enjoy this advantage while abandoning factivism, and without going all the way to the extreme internalist
15 One might insist that perceptual beliefs do warrant themselves. However, even if perceptual beliefs do warrant themselves, I take it perceptual states provide more warrant for beliefs than those beliefs provide for themselves. The challenge would then be to explain why perceptual states provide the greater degree of warrant they do. 16

My inspiration for this challenge comes from Martin (2001)s discussion of Brewer (1999). Martins challenge is to explain why, given Brewers emphasis on perceptual contents to explain perceptual warrant, states of imagination or hope are not likewise capable of providing warrant. A natural response to the challenge is to insist that perceptual states have a mind to world direction of t, whereas states of imagination or hopes do not. My present challenge to Burge aims to block this response, since states of belief will have a mind to world direction of t if any states do.

123

N. Silins

position mentioned in the introduction, on which the conscious character of experiences sufces to explain perceptual entitlement.17 Given that the factivist views are extraordinarily demanding, we should on balance prefer an approach which emphasizes consciousness and leaves the factivist views behind.

4 Conclusion Burges Perceptual Entitlement is a landmark account of how perceptual states give us warrant for beliefs. The account is distinctive in particular in the way it minimizes considerations about the conscious character of perceptual states, and in the way in which it foregrounds considerations about reliability and the individuation of the content of perceptual states. I hope to have separated a number of key claims of the overall position, and to have carefully evaluated Burges case in their favor. Although I think his arguments succeed for the claim that perceptual states provide a kind of entitlement that is basic, this claim is not distinctive of his position. It is afrmed by rival theorists such as Pollock, Huemer, and Pryor. As far as Burges distinctive theses about the role of reliability and anti-individualism are concerned, they have yet to be established. The claim that perceptual entitlement is basic can be comfortably combined with a wide range of alternative views about the role of reliability in entitlement. And even if it is agreed that perceptual states must have a kind of reliability grounded in their nature to give entitlement, rival accounts which bracket Burges anti-individualism will need to be blocked, such as those accounts which emphasize factive perceptual attitudes. By looking at Burges claims and arguments in detail, I have tried to improve our understanding of the motivations and options for the more general proper function approach. I have also promoted a quite different approach. This approach minimizes considerations about reliability and the individuation of the content of perceptual states, and instead foregrounds considerations about the conscious character of experience. We have seen prima facie reason to focus on consciousness, given that such an approach neatly avoids the challenge of explaining why perceptual beliefs do not warrant themselves. But a more extensive discussion of the approach will have to take place elsewhere.
Acknowledgments Thanks to an audience at the RSSS/ANU Consciousness at the Beach Conference of 2009, and thanks in particular to David Chalmers, Anna-Sara Malmgren, Ram Neta, Derk Pereboom, Declan Smithies, Daniel Stoljar, and an anonymous reviewer for Erkenntnis. Thanks especially to Tyler Burge for comments on an earlier draft.

References
Alston, W. P. (1989). Epistemic justication: Essays in the theory of knowledge. Ithaca: Cornell University Press. Bergmann, M. (2006). Justication without awareness. Oxford: Oxford University Press.
17

See Silins (forthcoming) for more detail about the role of consciousness in perceptual entitlement.

123

Explaining Perceptual Entitlement Brewer, B. (1999). Perception and reason. Oxford: Oxford University Press. Burge, T. (2003). Perceptual entitlement. Philosophy and Phenomenological Research, 67, 503548. Byrne, A. (2001). Intentionalism defended. Philosophical Review, 110, 199240. Casullo, A. (2007). What is entitlement? Acta Analytica, 22, 267279. Goldman, A. (2008). Reliability and immediate justication. In Q. Smith (Ed.), Epistemology: New essays. Oxford: Oxford University Press. Huemer, M. (2001). Skepticism and the veil of perception. Lanham, MD: Rowman and Littleeld. Huemer, M. (2007). Compassionate phenomenal conservatism. Philosophy and Phenomenological Research, 74, 3055. Jackson, A. (2011). Appearances, rationality, and justied belief. Philosophy and Phenomenological Research, 82, 564593. Johnston, M. (2006). Better than mere knowledge? In T. S. Gendler & J. Hawthorne (Eds.), Perceptual experience. Oxford: Oxford University Press. Malmgren, A. (2006). Is there a priori knowledge by testimony? Philosophical Review, 115, 119241. Martin, M. (2001). Epistemic openness and perceptual defeasibility. Philosophy and Phenomenological Research, 63, 441448. McDowell, J. (1995). Knowledge and the internal. Philosophy and Phenomenological Research, 55, 877893. Neta, R. (2010). Can a priori entitlement be preserved by testimony? In A. Haddock, A. Millar, & D. H. Pritchard (Eds.). Social epistemology. Oxford: Oxford University Press. Peacocke, C. (2004). The realm of reason. Oxford: Oxford University Press. Plantinga, A. (1993). Warrant and proper function. Oxford: Oxford University Press. Plantinga, A. (1996). Respondeo. In J. Kvanvig (Ed.), Warrant in contemporary epistemology. Lanham, MD: Rowman and Littleeld. Pollock, J. (1974). Knowledge and justication. Princeton: Princeton University Press. s, Pryor, J. (2000). The skeptic and the dogmatist. Nou 34, 517549. Pryor, J. (2001). Highlights of recent epistemology. British Journal for the Philosophy of Science, 52, 95124. Pryor, J. (2004). Whats wrong with Moores argument? Philosophical Issues, 14, 349378. Sawyer, S., & Majors, B. (2005). The epistemological argument for content externalism. Philosophical Perspectives, 19, 257280. Silins, N. (2008). Basic justication and the moorean response to the skeptic. Oxford studies in epistemology: volume 2. Oxford: Oxford University Press. Silins, N. (Forthcoming). Seeing through the veil of perception. Mind. Tye, M. (2000). Consciousness, color, and content. Cambridge: MIT Press. Wedgwood, R. (2002). Internalism explained. Philosophy and Phenomenological Research, 65, 349369.

123